A background-profile subtraction method improves zero-shot sound classification accuracy under noisy conditions, and narrowing the audio-text modality gap further boosts performance.
Apart from this, ATMs also face the challenge of themodality gap, wherein embeddings from differ- ent modalities are separated into distinct regions within the la- tent space
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.SD 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Domain Adaptation Method and Modality Gap Impact in Audio-Text Models for Prototypical Sound Classification
A background-profile subtraction method improves zero-shot sound classification accuracy under noisy conditions, and narrowing the audio-text modality gap further boosts performance.